OnionMHC

OnionMHC predicts peptide binding affinity to the human Major Histocompatibility Complex allele HLA-A*02:01 to support antigen presentation research and peptide therapeutic development.

Key Features:

• Integration of Structure and Sequence Data: Incorporates both structural information and amino acid sequence features to enhance predictive accuracy for peptide-MHC binding.
• Advanced Machine Learning Techniques: Employs natural language processing (NLP) to encode sequence information and convolutional neural networks (CNNs) within a deep-learning framework to extract patterns from structural data.
• Benchmark Performance: Evaluated on 18 weekly Immune Epitope Database (IEDB) benchmark datasets and experimentally validated peptides derived from whole-exome sequencing of breast cancer patients, showing improved performance relative to sequence-only models.

Scientific Applications:

• Peptide-Based Therapeutics: Predicts peptide-MHC binding affinities to inform design and optimization of therapeutic peptides.
• Cancer Vaccine Development: Screens potential neo-epitopes from whole-exome sequencing data to support personalized cancer vaccine development.
• Immunological Research: Enables investigation of antigen presentation and peptide-MHC interaction mechanisms.

Methodology:

Integrates structural and sequence-based features into a deep-learning framework, using NLP for sequence encoding and CNNs for extracting patterns from structural data.